akundaje
commented
2 months ago We will try to add this feature in the future (early next year) with a significant refactor as well. We currently don't have bandwidth to add this feature.
On Sun, Sep 15, 2024, 9:18 AM BartTheeuwes @.***> wrote:
Dear,
Thank you all for making this package. I'm currently trying to run ChromBPNet on the tutorial data. I have noticed that every time I rerun chrombpnet pipeline, I need to restart from scratch and delete the old chrombpnet_model folder. Because of this, it is regenerating the bigwig files from the Bam input every time. I expect that it should just be able to continue from a previously generated bigwig file, especially when training the same model with different folds... Is there a way to make the chrombpnet pipeline pick up from a specific point, or even just specifically from the training step?
All the best, Bart
— Reply to this email directly, view it on GitHub https://github.com/kundajelab/chrombpnet/issues/207, or unsubscribe https://github.com/notifications/unsubscribe-auth/AABDWEJNLQMIWNRKTNA3ODTZWWXNXAVCNFSM6AAAAABOH6JBBKVHI2DSMVQWIX3LMV43ASLTON2WKOZSGUZDOMBQGI3TCMI . You are receiving this because you are subscribed to this thread.Message ID: @.***>
Dear,
Thank you all for making this package. I'm currently trying to run ChromBPNet on the tutorial data. I have noticed that every time I rerun chrombpnet pipeline, I need to restart from scratch and delete the old chrombpnet_model folder. Because of this, it is regenerating the bigwig files from the Bam input every time. I expect that it should just be able to continue from a previously generated bigwig file, especially when training the same model with different folds... Is there a way to make the chrombpnet pipeline pick up from a specific point, or even just specifically from the training step?
All the best, Bart